Prevalence of bipolar disorder in multiple sclerosis: a systematic review and meta-analysis

Abstract

Background

Multiple sclerosis (MS) is a chronic disabling, demyelinating disease of the central nervous system and is often associated with psychiatric comorbidities. Some studies suggest increased prevalence of bipolar disorder (BD) in MS.

Objective

To conduct a systematic review and meta-analysis assessing the prevalence of BD in adults with MS.

Methods

We registered this review with PROSPERO and searched electronic databases (Ovid MEDLINE, Central, Embase, PsycINFO and Scopus) for eligible studies from earliest inception to October 2020. Prevalence data of BD in adult patients with MS were extracted. Meta-analysis was conducted using random-effects model.

Findings

Of the 802 articles that were screened, 23 studies enrolling a total of 68 796 patients were included in the systematic review and meta-analysis. The pooled prevalence rate of BD in patients with MS was 2.95% (95% CI 2.12% to 4.09%) with higher prevalence in the Americas versus Europe. The lifetime prevalence of BD was 8.4% in patients with MS. Subgroup analysis showed a higher prevalence of BD in MS in females (7.03%) than in males (5.64%), which did not reach statistical significance (p=0.53).

Conclusions

This meta-analysis suggests a high lifetime prevalence of BD in patients with MS. Patients with MS should be routinely screened for BD. Further assessment of bipolar comorbidity in MS through prospective studies may help in developing effective management strategies and may improve treatment outcomes in patients with MS.

Background

Multiple sclerosis (MS), a chronic demyelinating disease of the central nervous system and is associated with a high rate of depression and cognitive dysfunction.¹ With a global median prevalence of 33 per 100 000 people, MS is estimated to affect over 2 million people.² Its prevalence varies greatly based on the geographical location, being highest in North America and Europe (164.6 and 127 per 100 000, respectively).^{2 3} MS is seen more frequently in women than in men, with a female to male ratio ranging from 2:1 to 3:1.^3–5 However, men presenting with the disease may demonstrate a more severe decline in their cognitive ability.⁶ MS is one of the leading causes of neurological disability among young and middle-aged adults.⁷ Although people can be diagnosed at any age, the average age of MS onset is 30 years and may progress to irreversible disability by an average age of 45 years.⁸

Several medical and psychiatric comorbidities are associated with MS and are often linked to delayed diagnosis, an increase in the conversion from clinically isolated syndrome to MS, an increase in the rate of conversion from relapsing to progressive disease and diminished quality of life.^{9 10} Among the neuropsychiatric comorbidities in MS, mood disorders are the most commonly reported mental health condition.¹¹ Bipolar disorder (BD) is characterised by abrupt mood swings with fluctuating mood states and energy. BD is a severe mental illness with significant cognitive and functional impairment, reduction in quality of life and increased mortality.¹² BD pose a major public health burden, with an estimated life-time (and 12-month) prevalence rates of 1.0% (0.6%) for bipolar I disorder (BP-I), 1.1% (0.8%) for bipolar II disorder (BP-II) and 2.4% (1.4%) for subthreshold BD among the US adults.^{13 14} Patients with BP-I have a prominent history of intense mania with fluctuating mood lasting for 7 or more days or inpatient hospitalisation regardless of the duration. Patients with BP-II on the other have lower intensity hypomania symptoms lasting for 4 or more days and more prominent depressive symptoms. Prevalences of BD among adults are overall similar in both men and women,¹⁴ while women appear to be more prone to BP-II, with predominant depressive episodes and higher burden of medical comorbidities.^{15 16} The link between MS and BD has not been fully established, although a possible genetic association involving the HLA region has been suggested in patients with MS and a family history of BD.¹⁷

Objective

We conducted this systematic review and meta-analysis to explore the epidemiological coherence between BD and MS by the evaluating the prevalence of BD in adult MS population.

Study selection and analysis

Registration and search strategy

The study protocol was registered with the PROSPERO international prospective register of systematic reviews (http:// www.crd.york.ac.uk/prospero), registration ID: CRD42018104068. A comprehensive electronic search of academic databases (Ovid MEDLINE Epub Ahead of Print, In-Process & Other Non-Indexed Citations, Ovid MEDLINE Daily and Ovid MEDLINE, Central, Embase, PsycINFO and Scopus) from inception of each database to 15 October 2020, was designed and conducted by experienced librarians (online supplemental eTable 1: Complete search strategy). We included studies of any language and with adult participants. No restrictions were made with regard to the year of publication or the language used. Case reports, case series, intervention or treatment studies were excluded. This study is reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (http://www.prisma-statement.org).

Study selection and data extraction

Three investigators (BJ, ALN and NG) independently reviewed the studies by title, abstract and full text where relevant for inclusion. Studies were relevant if they reported data on prevalence of BD. Disagreements were resolved by consensus or by consulting a fourth investigator (BS). Articles were excluded if they did not report original study data (eg, reviews, book chapters), evaluated child/adolescent population prevalence, or did not report sufficient data. If there were multiple papers from the same location or multiple studies including the same population, only the article with the largest number of participants was used.

Relevant data were extracted using a piloted electronic data extraction form modified to fit the parameters of interest. Extracted data included author details (names and year of publication); study characteristics (country/region, setting, data source, sampling method, sample size, data collection period); demographics and disease characteristics (diagnostic criteria used, prevalence). For articles in which key data were missing, study authors were contacted.

Methodological quality assessment

Methodological quality assessment was performed using the tool developed by Hoy et al for prevalence studies.¹⁸ This tool consists of 10 items assessing the study quality and a summary risk of bias assessment. Assessment items include study sample’s representation of national population, sampling frame, sample selection, non-response bias, data source, case definition, measurement instrument, data collection, prevalence period and data reported. In all, four studies were categorised as having a low risk while the remainder had a moderate risk (online supplemental eTable 2: Quality assessment).

Data synthesis

Statistical analyses were performed using R software for statistical computing (V.3.6.1, www.R-project.org) with ‘meta’ (V.4.9.6) and ‘metafor’ (V.2.1.0) packages in RStudio (V.1.2.1335; www.rstudio.com). A DerSimonian and Laird random-effects model was used to estimate the summary effect size and its corresponding 95% CI from logit transformed observed proportions. Heterogeneity was quantified using the I² statistic and was further explored through moderator analysis. Outlying and influential studies were identified using a set of case deletion diagnostics proposed by Viechtbauer and Cheung,¹⁹ and the influence of each study on the summary proportion was estimated using leave-one-out analyses. Subgroup analyses were performed by: regions (Americas vs Europe; Europe vs USA), study setting (clinic vs database), reported prevalence type (point vs lifetime), BD type (BD-I vs BD-II), study type (clinical vs epidemiological vs database) and gender (female vs male).

Findings

Study characteristics

Of all the 802 non-duplicate papers screened, a total of 23 studies were deemed eligible for systematic review^20–42 (figure 1). All the 23 studies, with a total of 68 796 participants, were included in the meta-analysis. Eighteen studies reported the prevalence rate of BD in MS and five studies reported lifetime prevalence of BD in MS. The mean (SD) participant age from the 15 studies providing this information was 44.1 (13.5) years. The proportion of female participants in the study sample was reported in 17 studies, with a total of 41 067 female participants among 57 711 participants (71.16%). Of the 23 studies included in the meta-analysis, a total of 11 studies were from USA, and 11 studies were from Europe and one study was from Asia. An overview of the study characteristics is presented in table 1.

Figure 1.

Flow chart.

Table 1.

Study characteristics

Sl. no	Author, year	Location	Diagnostic criteria/tools	Study (population)/selection method	N (female %)	Age (mean, SD)	Prevalence/association of BD in MS
1	de Cerqueira, 201520	Brazil	McDonalds Criteria for MS, Expanded Disability Status Scale (EDSS), Beck Depression Inventory (BDI), Mini International Neuropsychiatric Interview (V.5) for DSM-IV, Beck Inventory for Anxiety	Cross-sectional study between January 2012 and December 2013. Convenience sample of consecutive patients with MS (age 19–65) from a public university-based outpatient service for Neuroimmunology in Rio de Janeiro, Brazil.	60 (76.7)	43–11.8	Prevalence of BD: 8/60
2	Demakis, 200921	USA	Physician-documented diagnosis coded in the data base.	Admission assessments for year 2000 from minimum data set, a federally mandated national database, of all residents in Medicare and Medicaid certified nursing facilities/nursing home residents in the USA.	924 (72)	57.5–13.5	Prevalence of BD: 21/924
3	Edwards, 200422	UK	Outpatient clinic database.	Consecutive outpatients (18–80 years) attending a MS clinic at University Hospital, Nottingham, between June 2002 and June 2003	658 (69)	45, NA	Prevalence of BD: 02/658
4	Espinola-Nadurille, 201023	Mexico	DSM-IV, Structured Clinical Interview for DSM-IV (SCID-I), MADRS, HAM-D, HAM-A, EDSS.	Consecutive patients who fulfilled McDonald criteria for the diagnosis of MS after the evaluation by two neurologists.	37 (64.8)	36.3–11.5	Prevalence of BD: 06/37
5	Feinstein, 200124	Canada	SCID-1, Pathological Laughing and Crying Scale, BDI, 28 item General Health Questionnaire.	A consecutive sample of 100 patients with clinically definite MS.	100 (69)	44.6–11.4	Prevalence of BD: 0/100
6	Galeazzi, 200525	Italy	SCID- I, BDI, State-Trait Anxiety Inventory (STAI)	50 consecutive patients with clinical relapsing–remitting MS	50 (52)	34.9–9	Lifetime prevalence of BD: 03/50
7	Gerber, 201726	Canada	Health administrative database using ICD-10-CA and ICD-9-CM	Patientswith MS (aged 35–55) started on disease-modifying therapy during 1 April 2011 to 31 March 2014, from Alberta health system administrative data.	2864 (73.8)	NA	Prevalence of BD: 136/2864
8	Horton, 201027	Canada and USA	Self-administered questionnaire and review of medical records	Patients18 years or older, followed at the provincial MS Clinic at the Health Sciences Centre in Winnipeg, Man., Canada and the MS Clinic (Mellen Centre) at the Cleveland Clinic in Cleveland, Ohio, USA	404 (76)	46.5–11.8	Prevalence of BD: 03/404
9	Joffe, 198728	Canada	Schizophrenia-Lifetime Version, BDI, Spielberger State Anxiety Inventory, Symptom Checklist-90 Revised. Psychiatric diagnoses were determined independently for each patient by two investigators.	Consecutive 100 patients of age range 20–71, who kept their regular appointment at the St. Michael’s hospital, Toronto	100 (63)	42.3–13	Lifetime prevalence of BD: 13/100
10	Jun-O' Connell, 201729	USA	Multiple Sclerosis Quality of Life-54 (MSQOL-54), Mood Disorder Questionnaire, SCID-I/NP by psychiatry trained physicians	Consecutive patients with a diagnosis of MS by 2010 revised McDonald Criteria and an age of between 18 and 90 years old were enrolled between January 2014 and May 2015.	152 (75)	49, NA	Prevalence of BD: 10/152
11	Kotan, 201930	Turkey	Neurological examinations were done by a neurologist. SCID-1-TR,Hospital Anxiety and Depression Scale, MSQOL-54, Psychosocial Adjustment to Illness Scale-Self Report (PAIS-SR), Quality of Life Scale Short Form 36, Multidimensional Scale of Perceived Social Support.	Patients who applied to Neurology department and were diagnosed as MS were recruited	227 (71.8)	37–9.9	Prevalence of BD: 4/227
12	Kurnaz, 201931	Turkey	Data from Medical records evaluated retrospectively.	126patients with MS with no comorbidities at MS onset	126 (70.6)	43.1–12.6	Prevalence of BD: 4/126
13	Laroni, 201732	Italy	McDonald criteria, EDSS, Patient databases	Newly diagnosed patients with MS enrolled longitudinally in 20 MS centres in Italy with a diagnosis of MS since 2010.	1877 (64.9)	35.3–11.3	Prevalence of BD: 11/1877
14	Lo fermo, 201033	Italy	Medical records data, DSM-IV-TR, EDSS	Retrospective review of medical records of patients with MS during the period 1997–2007 at University of Catania’s MS Centre	682 (NA)	NA	Prevalence of BD: 1/682
15	Lorefice, 201534	Italy	Poser’s classification, McDonald criteria, DSM-IV criteria, Advanced Neuropsychiatric Tools and Assessment Schedule (ANTAS), semi structured clinical interview derived in part from the nonpatient version (SCID-I/NP) for DSM-IV, (ANTAS-SCID), brief repeatable battery of neuropsychological tests	Outpatients affected by MS, consecutively referred to the MS Centres of the University of Cagliari, Sardinia	240 (69.6)	40.6–9.8	Prevalence of BD: 36/240
16	Marrie, 200936	USA	North American Research Committee on Multiple Sclerosis (NARCOMS), a self-report registry for patients with MS.	Responders to the NARCOMS Fall 2006 Update questionnaire	8828 (NA)	NA	Prevalence of BD: 213/8828
17	Marrie, 201535	Canada	Diagnosis codes, ICD-9 or ICD-10-CA system from administrative health database.	Data from 4 Canadian provinces-Canadian provinces, British Columbia, Manitoba, Quebec, and Nova Scotia in 2005	44 452 (71.3)	43.8–13.7	Prevalence of BD: 2089/44452
18	Passarell, 201737	Spain	Administrative health database	Population-based administrative health data of adult patients with MS from Catalonia	5548 (69.7)	48.26–12.73	Prevalence of BD: 45/5548
19	Ron, 198938	UK	Clinical Interview Schedule, BDI, Social Stress and Support Interview, Mini Mental State Exam, DSM-III-R	116patients with a definite MS diagnosis (age 23–67) from a larger pool of consecutive attendees for an MRI study in MS at the National Hospital for Nervous Diseases	116 (68.1)	38.8, NA	Lifetime prevalence of BD: 8/116
20	Uca, 201639	Turkey	McDonalds Criteria for MS, DSM-IV-SCID-I/CV, DSM-III-R-SCID-II, EDSS	55 relapsing-remitting patients with MS admitted to Clinic of Neurology, Konya	55 (85.5)	34.07–8.16	Prevalence of BD: 0/55
21	Uguz, 200840	Turkey	DSM-IV (SCID-I), EDSS	Consecutive patients (18–65 years of age) from MS Outpatient Clinic between February 2005 and October 2005.	74 (67.6)	34.57–11.93	Prevalence of BD: 1/74
22	Wang, 201841	Taiwan	ICD-9-CM codes from database	Patients with systemic autoimmune diseases from The National Health Insurance Research Database 2000–2011	1060 (NA)	NA	Prevalence of BD: 4/1060
23	Weaver, 201942	Chile	Psychological profile, Outcome Rating Scale, and Session Rating Scale following individual and group sessions	Patients with MS who were at risk of developing emotional distress refered by their neurologists	162 (70)	NA	Bipolar disorder: 4%

DSM, Diagnostic and Statistical Manual of Mental Disorders; HAM-A, Hamilton Anxiety Rating Scale; HAM-D, Hamilton Depression Rating Scale; ICD-10-CA, International Classification of Diseases, Tenth Revision, enhanced Canadian version; ICD-10-CM, International Classification of Diseases, Tenth Revision, Clinical Modification; MADRS, Montgomery–Åsberg Depression Rating Scale; MS, multiple sclerosis; NA, not available.

Prevalence of BD in MS

The overall pooled estimate of crude prevalence of BD in MS using random-effects meta-analysis was 2.95% (95% CI 2.12% to 4.09%) (figure 2). Substantial heterogeneity was found in the analysis (I²=95%; p<0.01). Subgroup analysis was performed in an attempt to explore the heterogeneity. Moderator analysis did not show any significant difference between study setting (clinic vs database, p=0.12). A multivariate moderator analysis showed variables ‘prevalence type’, ‘study type’ and ‘study region’ contributed to 58.24% of heterogeneity (p<0.0001). Contribution of each study towards the Q-statistic for heterogeneity is visualised using a Baujat plot (online supplemental efigure 1).

Figure 2.

Forest plot—overall prevalence of BD in MS. BD, bipolar disorder; MS, multiple sclerosis.

Estimated pooled effect size for lifetime prevalence (based on five studies25 28 31 34 38 was 8.42%; 95% CI 4.50% to 15.21%); and for point prevalence, the rate was 2.13%; 95% CI 1.48% to 3.07%. Results from subgroup analyses showed a higher prevalence of BD in the Americas as compared with Europe (4.70%; 95% CI 3.04% to 7.19% vs 1.99%; 95% CI 1.20% to 3.28%; p=0.01). Data from three studies^{28 29 34} showed that the prevalence of BD among female patients with MS was higher than observed in males (7.03%; 95% CI 4.51% to 10.80% vs 5.64%; 95% CI 3.30% to 9.48%), but not significant (p=0.53). Four studies20 25 29 34 reported the proportions of BD per type in patients with MS. Subgroup analysis from these studies showed a higher prevalence of BD-II compared with BD-I (5.52%; 95% CI 2.38% to 12.29% vs 2.81%; 95% CI: 1.13% to 6.79%), but we did not find a statistically significant difference between groups (p=0.27), (online supplemental efigure 2).

The leave-one-out sensitivity analysis showed that no single study significantly influenced the analysis (figure 3). Overall summary effect size after excluding outlier was 3.31%; 95% CI 2.44% to 4.48% (online supplemental efigure 3). Sensitivity analysis performed with Brenner et al,⁴³ a large study that was not included as it did not meet the inclusion criteria. This study reported a 0.5% prevalence of BD in cohort of 10 750 patients with MS identified using the Swedish national patient register. The sensitivity analysis showed a pooled prevalence of 2.59%; 95% CI 1.78% to 3.75% (online supplemental efigure 4). A sensitivity analysis was also performed using Freeman-Turkey transformation, and the overall pooled prevalence (2.88%; 95% CI1.84% to 4.13%) did not significantly differ from logit transformation. Subgroup analysis based on the study type (epidemiological, clinical, and database) is reported in online supplemental efigure 5. Egger’s test showed a significant asymmetry in funnel plot (z=3.1538, p=0.0016). Publication bias is visualised using trim and fill funnel plot (figure 4).

Figure 3.

Forest plot—leave one out sensitivity analysis. The reference line indicates where the original summary proportion lies. Each box represents a summary proportion estimated leaving out a study.

Figure 4.

Trim and fill funnel plot.

Discussion

This comprehensive systematic review and meta-analysis evaluated the prevalence of BD in adult populations with MS, with overall crude prevalence rate of almost 3%. The prevalence of BD among women (although it was higher) was not statistically significantly different than men in patients with MS. The pooled lifetime prevalence of BD in MS is approximately 8.4%. Although the lifetime prevalence of BD in MS is much higher, there were only five studies which reported lifetime prevalence data, had smaller size and were conducted mostly at the referral centres. Thus, a possibility of referral bias could not be completely ruled out.

Studies from Europe reported a lower prevalence of BD in MS when compared with studies from the Americas. Three studies (N=9904) from USA showed a pooled prevalence of 3.29% (95% CI 0.76% to 13.04%) (online supplemental efigure 2). These findings are consistent with the epidemiological data showing the higher prevalence rate of BD in the USA as compared with some of the European countries.⁴⁴ The reason of difference in the prevalence rates could largely be due to study design, assessment differences and genetic vulnerability. Recent data has suggested a significantly earlier age of onset of BD in the USA, more genetic and psychosocial stressor vulnerability and comorbidities among patients living in the USA as compared with patients living in Europe.⁴⁵ The estimated prevalence of BD in MS is larger than reported in a recent meta-analyses of BD prevalence in general population.⁴⁶ However, our study cannot answer whether BD is more prevalent in MS than the general population. To address this important question, studies reporting on the association between BD and MS compared with the general population should be considered.

Patients with MS are known to have white matter abnormalities both in areas of known white matter lesions and in normal appearing white matter.⁴⁷ BD is associated with white matter diffusion abnormalities, predominantly in the right posterior temporoparietal and left cingulate regions.^{48 49} Cortical demyelination is a hallmark of MS, and is typically not detected by standard brain MRI.⁵⁰ Patients with BD have been demonstrated to have thinning of the cortical grey matter in the frontal, temporal and parietal regions of both hemispheres.^{49 51} MS lesions in critical brain areas could be a substantial contributing factor towards an increased diagnosis of BD in this population. However, an unambiguous link between the location of these lesions and psychiatric manifestations is yet to be conclusively demonstrated. BD is known to be associated with immune dysfunction, and at least in a subset of BD, it plays a significant role in the pathophysiology of the disease.⁵² Both MS and BD are HLA linked, but not to the same HLA haplotypes.^{17 53 54} The presence of mood disorders in MS is associated with an increased risk of death, diagnostic delay of MS, an increased risk of MS disease activity, an increased risk of progression and increased risk of MS associated disability.^{55 56}

There are several limitations to this comprehensive systematic review. BD is a complex disorder and the current method of diagnosing BD as a construct may contribute to diagnostic errors including misdiagnosis, underdiagnosis and sometimes overdiagnosis.^57–59 Studies included in this review used different methodologies for arriving at the diagnosis of BD. Although most studies included in this review had a diagnosis of BD made by trained personnel, structured interviews or by retrospective chart reviews, some studies employed less rigorous methods such as patient self-report. This could have contributed to the higher prevalence rate observed in this study. Only one study reported BD diagnosis using self-report. We conducted a leave-one-out sensitivity analysis (figure 3) by omitting one study at a time to examine the influence of each study on the pooled estimates, and we did not find a significant difference in the pooled estimates. This study did not evaluate the effects of MS treatment as a potential risk factor for development of BD. We also did not assess the longitudinal course of MS disease progression to see if there were more cases of BD diagnosed during later stages of MS. BD has a 5–10 years mean delay in diagnosis from the first episode of illness.¹² We were unable to determine if the increased healthcare utilisation following an MS diagnosis has improved the chances at diagnosing BD in this study cohort. Only two studies included in this analysis looked at the prevalence of BD at MS diagnosis.^{32 33} Most studies included for data extraction were specialised clinic-based studies and were of moderate methodological quality. Studies also differed in terms of the assessment criteria used to diagnose BD. The database-based studies included had several limitations, including inadequate data on the diagnostic criteria used and lack of data verification. Statistical tests suggested the presence of publication bias; although such tests have not been designed for prevalence studies. Studies with large sample sizes or statistically significant findings are more likely to be published and therefore captured in a systematic review. Publication bias can exaggerate the estimates of the main effect and may also underestimate the variance.⁶⁰ All these factors together might have contributed to the increased variability and heterogeneity across the studies in the current meta-analysis. We also had very limited to no data from Asia, Africa and Australasia.

This systematic review has several strengths. It followed the standard methods of systematic reviews and meta-analysis including the Cochrane handbook and PRISMA statement. We performed an extensive database search using reproducible criteria to capture the most number of studies on BD prevalence in patients with MS worldwide. We used robust statistical methods to pool studies while accounting for the effects of outliers. We also investigated potential sources of heterogeneity.

Conclusions and clinical implications

In summary, this meta-analysis provides evidence that there is an increased prevalence of BD among patients with MS. Further assessments of bipolar comorbidity in MS through prospective studies may help in developing effective management strategies and improve treatment outcomes. Future studies are necessary to evaluate the potential mediators and risk factors of BD in patients with MS, including the roles of structural brain abnormalities, inflammation and MS treatment regimens. Early identification of BD can help increase treatment adherence and improve the quality of life in patients with MS.

Data availability statement

Data are available on reasonable request. The data that support the findings of this study are available from the corresponding author on reasonable request.